Are all Polls Equal? Analyzing the Polls of the US2020 election, a new Perspective

Claire Durand
Additional contact information
Claire Durand: Université de Montréal

No h87vq, OSF Preprints from Center for Open Science

Abstract: At the dawn of the 2024 American presidential campaign, it is not pointless to reassess what happened during the 2020 campaign. In that election, the polls have been the least accurate since 1996, with a notable disparity in results depending on the poll's mode of administration and sampling frame or source. Based on their methodology, the 222 national-level campaign polls were categorized as mixed-mode (16%), single-mode quasi-random polls (25%), and web opt-in polls (59%). Using local regression and multilevel analysis, the study revealed differences in campaign trends across categories. All the polls using random or quasi-random sampling indicated an initial rise in voting intention for Joe Biden followed by a decline until election day, while web opt-in polls showed his support as stable. Notably, mixed-mode polls provided an almost perfect election forecast. The poll estimates of the last ten days support these findings, showing higher accuracy within the mixed-mode and single-mode quasi-random polls compared to web opt-in polls. The study suggests that different modes and sources capture varying segments of the population, leading to more accurate polling. The results stress the need for academia, media, and pollsters to closely monitor the methodological diversification introduced in the 2020 election.

Date: 2024-03-07
New Economics Papers: this item is included in nep-pol
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://osf.io/download/65f0aa30e5e51c0e67bc593f/

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:osf:osfxxx:h87vq

DOI: 10.31219/osf.io/h87vq

Access Statistics for this paper

More papers in OSF Preprints from Center for Open Science
Bibliographic data for series maintained by OSF ().